A 50VA transformer will take about 0.21 Amps when correctly loaded (VA / Input voltage)
So a fuse of about 1.5 x the input is suggested and should be Anti Surge (Usually marked T or TT - T stands for "träge" which is german for Lazy or slow) - so 315mA A/S or 400mA A/S
If your fuses are vaporized and cover insides with remains of the wire - This indicates a major short...
What type of transformer are you using - is it a toroidal - if so have you got a shorted turn (if you mount a toroidal transformer incorrectly you can add an extra winding which is shorted out - this is creates by mounting the transformer with a conductive clamp which is bolted down in the middle - if you are using a toroidal - try removing the clamp...)
It is possible that you have a faulty diode in your bridge - I have seen diodes that measure OK when tested with a meter, but when either loaded, or subjected to a higher voltage, break down and become shorts, or leak - the easiest way to prove is to replace ALL the diodes, as I have found if one is faulty, it usually subjects others in the bridge to stress, which may make them more likely fail, and for the cost of 4 diodes of 1N400X or 1N540X - I usually use 1N4007 or 1N5408...
The Raspberry Pi is generally powered from a 5V wall-wart type of DC power supply. The secondary is galvanically isolated from the mains voltage for reasons of personal safety (a fault will not expose the end user to the mains voltage).
The DC return of your bridge rectifier circuit is most certainly not isolated from the mains. The Raspberry Pi ground is 'floating' with respect to the bridge rectifier ground - there is no galvanic connection between them, hence your voltage measurement.
If you were to connect the DC return of the bridge rectifier circuit to the Raspberry Pi ground, you bypass the galvanic isolation that the DC power supply gives you. This means your Raspberry PI is now mains-referenced, and any fault could potentially expose you you to lethal voltages. I wouldn't do this.
A further complication comes if you also hook up an earth-referenced return to the Raspberry Pi, like a connection to a PC, with the mains-reference return connected. When you mix a mains-referenced return like your rectifier circuit with earth, things are going to explode (you essentially short out your bridge through the earthed return, which is often a flimsy wire that gets really hot and melts/catches fire while blowing up everything connected to it). Another reason not to do this.
You would be much better off with a small line frequency transformer to (1) step down the mains voltage to a lower level ahead of your resistive attenuator, and (2) provide galvanic isolation from the mains. Put your bridge and attenuator in the secondary of the transformer. With this, you can safely connect the low voltage isolated rectifier return to the Raspberry Pi return.
(You also must include a fuse in the line to isolate the rectifier circuit from the mains if there is a severe fault like a transformer fault or a short circuit.)
Best Answer
The voltages shown are measured relative to ground, and they are assuming an AC-coupled average-reading multimeter. It's a simple bridge rectifier. The 10nF capacitors are to reduce buzz in the audio from the diodes.
The AC voltage coming out of the transformer secondary (red-to-red wires) will be more like 240VAC RMS. Half appears at each side relative to ground (when your multimeter subtracts the DC component as they normally do).
Peak voltage from a bridge rectifier, of course, is \$\sqrt{2}\$ times the RMS voltage, and the filter capacitors mean that you will measure close to that (minus a bit due to the ripple under load and the diode drops).
Here is a simulation showing the two voltages AC1/AC2 relative to ground and the transformer output voltage (AC1-AC2) and the DC output voltage (relative to ground).
If you imagine an average line for AC1/AC2 in the middle between the two peaks and ignore that DC offset (as your meter will on an AC volts range unless it is not an ordinary multimeter) you can see that the total amplitude of each relative to ground is half of the transformer winding output voltage.